Interview: 'when should you NOT use RL?'
mediumAnswer
- (1) Supervised data + clear labels available — SL is faster + safer.
- (2) Task has one shot (no sequential decisions).
- (3) Reward hard to specify — SL on demos may be better.
- (4) Cannot simulate / afford exploration + no logged data.
- (5) Business needs interpretability RL lacks.
- (6) Costly failures + can't validate before deployment.
- Rule: RL is expensive to get right; use only when the sequential / decision structure genuinely warrants it.
Check yourself — multiple choice
- Always use RL
- Skip RL when: SL data available / one-shot / hard reward / no sim + no logs / need interpretability / can't validate — RL expensive, use only when sequential structure warrants it
- Random
- Never
Skip RL: SL sufficient / one-shot / hard reward / no sim / need interp / safety.
#interview
Practise Reinforcement Learning
214 interview questions in this topic.
Related questions
- Interview: 'design an RL system for X' — what's your framework?
- Interview: 'your DQN agent isn't learning — how do you debug?'
- Interview: 'explain why PPO is on-policy but uses a ratio to importance-sample.'
- Interview: 'walk me through the tradeoffs of SFT-only vs SFT+RLHF vs SFT+DPO.'
- Interview: 'you're given 100 hours of expert data + can simulate — how do you train an agent?'
- Interview: 'how do you make an RL policy safe to deploy?'